NVIDIA’s Tegra business
NVIDIA’s (NVDA) core products are GPUs (graphics processing units), but the demand for SoCs (system-on-chip) is increasing. This shift toward a system-based approach would gradually expand NVIDIA’s Tegra processor business. The company now earns 15% of its revenues from Tegra.
Its Tegra business consists of game consoles, self-driving cars, infotainment systems, and the Jetson platform for computing on the edge. The company doesn’t give earnings from each of the three sub-segments separately.
NVIDIA’s Tegra revenue rose 101% YoY (year-over-year) to $333 million in fiscal 2Q18. This growth was largely driven by strong demand from Japan-based (EWJ) Nintendo for its Switch gaming console. Nintendo Switch is powered by NVIDIA’s Tegra X1 ARM processor.
Nintendo sold ~2.0 million Switch units in calendar 2Q17, compared with 2.7 million units sold in calendar 1Q17. Despite the decline in Switch unit sales, NVIDIA’s Tegra revenue growth was sequentially flat as the decline in orders from Switch was offset by increasing demand from automotive and edge computing.
Tegra has strong growth potential as NVIDIA expands the scope of its AI (artificial intelligence) beyond data center and into every device. The biggest growth will likely come from autonomous cars.
NVIDIA earns only 12% of its revenue from automotive, wherein a majority of its revenues come from infotainment systems. Its self-driving platform, DRIVE PX 2, is still in the development stage. The company will start deploying the DRIVE PX 2 AI (artificial intelligence) car platform in Tesla’s (TSLA) and Toyota Motor’s (TM) cars by the end of fiscal 2018, which will likely increase Tegra’s revenues.
After autonomous vehicles, another major revolution will likely come from AI at the edge, wherein basic deep learning would happen inside the device instead of data center. The company’s Jetson platform would power smart cameras used for surveillance and smart traffic lights used to monitor traffic. The company is building a Metropolis Software Partner Program that would be at the core of the future AI cities.
The company expects its total addressable market in AI cities to reach $2 billion by 2020, or $1 billion from servers and $1 billion from devices on the edge. However, the AI city is still at the concept stage and will likely take time to materialize.